Acropolis

Muse Wonder Full 1.0

MetaGrade A. Evidence grade A of A to D: strongest.Weights unstatedReleased 12 Oct 2026Consistent across benchmarks
Compare with peers

Domain profile

Median within-board percentile by capability domain across boards with sufficient peer overlap.

  • Agentic (11 boards): 84 (scale 0 to 100)Agentic (11 boards)84
  • Coding (7 boards): 84 (scale 0 to 100)Coding (7 boards)84
  • Finance (4 boards): 85 (scale 0 to 100)Finance (4 boards)85
  • Legal (3 boards): 89 (scale 0 to 100)Legal (3 boards)89

Benchmark appearances

Best verified run per board, with confidence intervals and Pareto frontier status.

BenchmarkRankScoreCostStatusObserved
Finance Agent

Vals AI benchmarks · max

#4 / 38Tied #1
67.6%[63.7% to 71.5%]$2.68On frontier. Evidence grade A of A to D: strongest.15 Nov 2026
Legal Research

Vals AI benchmarks · max

#4 / 38Tied #1
53.4%[46.6% to 60.2%]$3.60On frontier. Evidence grade A of A to D: strongest.15 Nov 2026
Legal Agent Benchmark (HLAB-Vals)

Vals AI benchmarks · max

#5 / 38Tied #1
23.3%[16.4% to 30.3%]$5.41Standard15 Nov 2026
EMB

Vals AI benchmarks · max

#6 / 38Tied #1
75%[70.8% to 79.2%]$4.58Standard15 Nov 2026
Terminal-Bench 4

Vals AI benchmarks · max

#7 / 38Tied #1
58.1%[51.6% to 64.6%]$10.53Standard15 Nov 2026
Vals Index

Vals AI benchmarks · max

#7 / 38
66.8%[64.7% to 68.9%]$6.81On frontier. Evidence grade A of A to D: strongest.15 Nov 2026
LMArena WebDev

LMArena · Overall · max

#8 / 156
1,724[1,709 to 1,738]UnknownStandard15 Nov 2026
LMArena WebDev

LMArena · Webdev · max

#8 / 156
1,724[1,709 to 1,738]UnknownStandard15 Nov 2026
Tax Agent

Vals AI benchmarks · max

#8 / 38Tied #1
43%[36% to 50%]$2.15Standard15 Nov 2026
LMArena WebDev

LMArena · Webdev HTML · max

#9 / 155
1,723[1,682 to 1,765]UnknownStandard15 Nov 2026
LMArena WebDev

LMArena · Webdev react · max

#9 / 140
1,742[1,723 to 1,760]UnknownStandard15 Nov 2026
Code Migration

Vals AI benchmarks · max

#12 / 38
61.7%[53.1% to 70.2%]$23.24Standard15 Nov 2026
LMArena Agent

LMArena · Overall · max

#13 / 64
0.1[0.1 to 0.1]UnknownStandard15 Nov 2026
Vibe Code Bench

Vals AI benchmarks · max · OpenHands

#13 / 38
86%[82.2% to 89.8%]$2.29Standard15 Nov 2026
Bug Hunt Bench

Bug Hunt Bench · max · Muse Code

#16 / 145
42UnknownStandard14 Oct 2026
Bug Hunt Bench

Bug Hunt Bench · xhigh · Muse Code

#18 / 145
41.5UnknownStandard14 Oct 2026

Reasoning effort scaling

Same benchmark and harness measured across multiple reasoning effort settings.

  • Bug Hunt Bench

    Bug Hunt Bench · Muse Code

    Monotone. Evidence grade A of A to D: strongest.
    xhigh

    41.5

    max

    42.0

Closest head-to-head rivals

Direct pairwise win rates for Muse Wonder Full 1.0 across shared benchmark boards.